基于RQA与SVM的声发射信号检测识别方法  

Detection and identification of acoustic emission signals based on recurrence quantification analysis and support vector machines

作　　者：司莉[1,2] 毕贵红[3,2] 魏永刚[4,2] 陶然 张寿明[1] 

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500 [2]云南省特种设备安全检测工程技术研究中心,昆明650050 [3]昆明理工大学电力工程学院,昆明650500 [4]昆明理工大学冶金与能源工程学院,昆明650093

出　　处：《振动与冲击》2016年第2期97-103,123,共8页Journal of Vibration and Shock

摘　　要：针对裂纹声发射信号检测问题,提出基于递归定量分析与支持向量机相结合的新型检测方法。利用小波阈值去噪原理,对采集的声发射信号进行去噪,将递归定量分析引入声发射信号检测,提取递归定量分析的量化特征参数,结合支持向量机对模拟裂纹声发射信号进行识别。并实验验证该方法的可行性。To prevent the leakage accidents of pipes and boiler, the key technique lies in whether the signals of cracks and small leaks can be detected effectively. Aiming at this, a new method of detecting acoustic emission signals based on RQA and SVM was described. The theory of wavelet threshold de-noising was used to reduce the noise signal. The principle of RQA was borrowed to detect acoustic emission signals and by calculation, some quantifiable feature parameters were obtained. Using these parameters as the SVM input parameters, the simulated acoustic emission signals of cracks were identified. By experiments, the feasibility of the method was validated.

关 键 词：声发射信号 小波阈值去噪 递归定量分析 支持向量机 

分 类 号：TB52[理学—物理]